Taizhen Cheung

ONLINE // v.2026

ABOUT ME PLAYER STATS

Software Engineer Profile

I'm a software engineer and Computer Science + Mathematics student at Hunter College (CUNY), graduating May 2027. I build production-minded AI systems and full-stack applications across LLM evaluation, retrieval-augmented generation, machine-learning benchmarking, and data-intensive web products.

My work spans Python, TypeScript, React, Next.js, FastAPI, PostgreSQL/pgvector, PyTorch, and cloud deployment. I care about measurable quality: grounded outputs, reproducible evaluations, secure authentication, automated tests, and clear technical documentation.

1,000+ LLM PROMPTS BENCHMARKED

98.0% PRODUCTION RETRIEVAL TOP-1

207 RESUFILL TESTS

2027 EXPECTED GRADUATION

Hunter College, CUNY

CURRENT MAIN SF6 // MASTER RANK

SAGAT

Zoner // The King of Muay Thai

MASTERY 65% • PLATFORM PC • CFN PLAYER_ONE

Outside of code I main Sagat in Street Fighter 6 — grinding ranked, labbing frame data, and chasing that next rank-up. Try typing "hadouken" or "tiger" anywhere on this page.

EXPERIENCE LOG 4 MISSIONS LOGGED

Applied experience in LLM evaluation, human-feedback data quality, machine-learning research, benchmarking, system design, and the software development lifecycle.

AI Trainer – Software Engineering & Coding Domain

Handshake – Remote

Jun 2026 – Aug 2026

Evaluated frontier large language model responses across multi-turn software-engineering and coding tasks, scoring correctness, instruction following, helpfulness, and technical communication against detailed rubrics.
Authored evidence-based scoring rationales and comparative preference evaluations across multiple candidate responses to identical coding prompts.
Calibrated judgments to maintain high inter-rater agreement, producing high-quality human-feedback data used to improve the reasoning and response quality of AI coding assistants.

AI Trainer – Image Processing Domain

Handshake – Remote

Nov 2025 – May 2026

Used Python-based tooling to evaluate and validate multimodal training datasets, supporting high-precision labeling and consistent quality assurance for AI systems.
Applied structured annotation, rubric-based review, and dataset validation workflows to identify labeling inconsistencies and bias risks.
Contributed to image-processing and computer-vision datasets, documenting quality findings and validating training examples against project requirements.

Research Intern – LLM Bias and Benchmarking

TIER Lab, Hunter College – New York, NY

Jun 2025 – Sep 2025

Benchmarked 1,000+ prompts across Gemini 2.0 Flash, DeepSeek R1, and GPT-4o to measure reasoning consistency, response safety, latency, and model bias.
Designed bias audit protocols for LLM-integrated robotics systems, identifying critical failure modes in Human-Robot Interaction (HRI) safety and fairness.
Engineered reproducible Python benchmark datasets for robotic perception and dialogue evaluation across multiple model architectures.
Quantified latency and bias trade-offs between high-speed and reasoning-heavy models, delivering data-driven mitigation strategies to research stakeholders.

Capital One First-Gen Focus Fellow

Capital One

Oct 2024 – Dec 2024

Completed an eight-week development program with Capital One engineers, building foundational knowledge of system design, Agile development, and the software development lifecycle.
Strengthened software engineering knowledge through discussions with engineers on system design and development practices in financial services.

QUEST TIMELINE JOURNEY LOG

EXPERIENCE

PROJECTS

EDUCATION

TIP: Select any node to open its full mission log.

EXP

AI Trainer – Software Engineering & Coding Domain

Handshake – Remote

Jun 2026 – Aug 2026

Score frontier-LLM responses on multi-turn software-engineering and coding tasks against detailed rubrics, generating high-quality human-feedback data for AI coding assistants.

▸ VIEW DETAILS

EXP

AI Trainer – Image Processing Domain

Handshake – Remote

Nov 2025 – May 2026

Utilized Python-based tooling to evaluate and validate training datasets for LLMs. Streamlined data validation workflows and contributed to image domain dataset projects.

▸ VIEW DETAILS

PROJECT

Onward – AI Emotional-Companion Platform

Next.js 15, TypeScript, Supabase/pgvector, Cerebras GPT-OSS 120B, Gemini

May 2026 – Present

Built and evaluated versioned retrieval recipes; the production keyword + reranker gate reached 99/101 top-1 with zero definitive-wrong results.

▸ VIEW DETAILS

PAPER

When Directional Accuracy Lies

Sole-author preprint · arXiv:2607.12248v2 · Equity Forecasting

Mar 2026 – Jul 2026 · Preprint on arXiv

A base-rate-honest benchmark showing that strong directional accuracy can hide economically worthless equity predictions.

▸ VIEW DETAILS

PROJECT

xiaofinance – Investment-Sentiment Dashboard

Python, FastAPI, SQLite, Preact, TypeScript, MCP

Jul 2026

Local-first dashboard ranking US stocks and investment themes by 24-hour Xiaohongshu discussion volume, with verbatim-verified LLM sentiment summaries.

▸ VIEW DETAILS

PROJECT

Weather Forecast App

React, Node.js, Express.js, Vite, REST API

Jul 2025 – Aug 2025

Full-stack weather app with React frontend and Node.js/Express backend consuming OpenWeatherMap API for real-time forecasts.

▸ VIEW DETAILS

EXP

Research Intern – LLM Bias and Benchmarking

TIER Lab, Hunter College

Jun 2025 – Sep 2025

Comparative analysis of 1,000+ prompts across Gemini 2.0 Flash, DeepSeek R1, and GPT-4o. Designed bias audit protocols for LLM-integrated robotics.

▸ VIEW DETAILS

PROJECT

CppAutoSim – Autonomous Driving Simulator

C++, OOP, Algorithm Design

Feb 2025 – Mar 2025

Vehicle simulation engine with realistic kinematics, lane-change algorithms, and obstacle avoidance. Modular OOP architecture.

▸ VIEW DETAILS

EXP

Capital One First-Gen Focus Fellow

Capital One

Oct 2024 – Dec 2024

Completed an eight-week program with Capital One engineers, learning system design, Agile development, and SDLC fundamentals.

▸ VIEW DETAILS

EDU

Hunter College, CUNY

B.A. – Double Major: Computer Science & Mathematics

Expected May 2027

Coursework: Data Structures, Algorithms, Software Analysis & Design, Numerical Methods, Linear Algebra, Discrete Math

▸ VIEW DETAILS

PROJECT ARCHIVE 5 COMPLETE // 1 ACTIVE

Selected work across retrieval-augmented generation, ML evaluation, developer tooling, data pipelines, and full-stack engineering. Every public claim below is backed by source, tests, or committed results.

ARXIV PREPRINT

When Directional Accuracy Lies

Mar 2026 – Jul 2026 · arXiv:2607.12248v2

Authored and posted a sole-author preprint introducing a base-rate-honest benchmark for LoRA-adapted TimesFM equity forecasting.
Built leakage-controlled, expanding walk-forward evaluation across frozen S&P 500 and NASDAQ-100 universes with honest always-up, random-walk, persistence, AR(1), and zero-shot baselines.
Released checksum-verified data, committed results, reproducibility tooling, and 55 CPU tests.

Python PyTorch TimesFM LoRA Time-Series Financial ML

▶ READ PAPER ▶ VIEW REPO

Onward – AI Emotional-Companion Platform

May 2026 – Present

Built and deployed a Next.js 15 + TypeScript application with server-sent-event story streaming and a client-side story player state machine.
Engineered six-lane FacetsRAG as a versioned shadow challenger using vector search, reciprocal-rank fusion, and LLM reranking; retained the stronger keyword + reranker production recipe at 99/101 top-1 with zero definitive-wrong results.
Implemented provider-pluggable LLMs, Supabase Auth, session-ownership authorization, and atomic per-user/per-IP PostgreSQL rate limiting.

Next.js 15 TypeScript Supabase/pgvector Cerebras GPT-OSS 120B Gemini Vercel

▶ LAUNCH APP

DEVELOPER TOOL

resufill – Grounded Resume Tailoring Engine

Aug 2026

Built a local Python CLI and Preact + TypeScript interface that turns a canonical career record into job-specific, ATS-parseable resumes and cover letters.
Designed an evidence-grounding gate that traces every bullet to a source ID, rejects unsupported claims, and reports missing job-description keywords as honest gaps.
Shipped claim-by-claim audits, Chromium PDF rendering and parse-back verification, local FastAPI serving, and 207 tests.

Python FastAPI Pydantic Preact TypeScript Playwright

▶ VIEW REPO

xiaofinance – Investment-Sentiment Dashboard

Jul 2026

Built a local-first FastAPI + SQLite pipeline for ingesting, deduplicating, scoring, and analyzing 24-hour Xiaohongshu investment discussions.
Created a Chinese alias and retail-slang dictionary with investment-context gating; collapsed repost spam using simhash clustering.
Grounded LLM sentiment summaries by validating every rendered quote against its source and exposed the corpus through an MCP server for agent analysis.

Python FastAPI SQLite Preact TypeScript MCP

▶ VIEW REPO

Weather Forecast App

Jul 2025 – Aug 2025

Built a React + Node.js/Express full-stack application consuming OpenWeatherMap REST APIs for current conditions, geolocation, and five-day forecasts.
Structured the backend with controller/service layers, CORS, validation, and protected API keys; implemented Context API, custom hooks, routing, and Axios data services on the frontend.

React Node.js Express.js Vite REST API

▶ VIEW REPO

CppAutoSim

Feb 2025 – Mar 2025

Built a C++ autonomous-driving simulator with acceleration, braking, turning, lane changes, vehicle following, parking, and crosswalk behavior.
Implemented static-obstacle avoidance, overtaking, and oncoming-traffic scenarios through a modular object-oriented scene and planning architecture.

C++ OOP Algorithm Design

▶ VIEW REPO

▶ MORE ON GITHUB

TECH STACK SKILL INVENTORY

START A CONVERSATION PLAYER TWO?

Let’s build something reliable.

I’m interested in software engineering, AI/ML engineering, and applied research opportunities where careful evaluation and strong product execution matter.

▶ EMAIL ME ▶ LINKEDIN ▶ GITHUB ▶ RESUME PDF